[Sheepdog] [PATCH 00/18] support sheepdog shutdown and restart

MORITA Kazutaka morita.kazutaka at lab.ntt.co.jp
Thu Mar 11 07:47:59 CET 2010

This patchset supports cluster shutdown and restart. Sheepdog
automatically start only when it is safe. Otherwise you need to run
`shepherd mkfs` or fix manually.

The cases sheepdog doesn't start automatically are:
 - There is no local object (before running `shepherd mkfs`)
 - All nodes which were in the sheepdog cannot start up
 - There is a node which has wrong epoch info.
   e.g. mkfs is invoked at another time
 - Objects are updated after network had partitioned
 - etc..

When you want to know current cluster status, run the following

$ shepherd info -t cluster

Ctime              Epoch Id               Nodes
10-03-11 11:56:55      1 c114a7fc04c240b4 [,,]

This patchset is also in the next branch.


Kazutaka Morita

More information about the sheepdog mailing list